US20140006466A1 - High speed and low power circuit structure for barrel shifter - Google Patents
High speed and low power circuit structure for barrel shifter Download PDFInfo
- Publication number
- US20140006466A1 US20140006466A1 US13/538,001 US201213538001A US2014006466A1 US 20140006466 A1 US20140006466 A1 US 20140006466A1 US 201213538001 A US201213538001 A US 201213538001A US 2014006466 A1 US2014006466 A1 US 2014006466A1
- Authority
- US
- United States
- Prior art keywords
- decoder
- input
- shift
- signal
- shifted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F5/00—Methods or arrangements for data conversion without changing the order or content of the data handled
- G06F5/01—Methods or arrangements for data conversion without changing the order or content of the data handled for shifting, e.g. justifying, scaling, normalising
- G06F5/015—Methods or arrangements for data conversion without changing the order or content of the data handled for shifting, e.g. justifying, scaling, normalising having at least two separately controlled shifting levels, e.g. using shifting matrices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/76—Arrangements for rearranging, permuting or selecting data according to predetermined rules, independently of the content of the data
Definitions
- the present invention generally relates to logic circuits for computational devices, and more particularly to a barrel shifter or rotator.
- a shifter or rotator, is a combinational logic circuit that can shift an operand data value by a specified number of places (bits), i.e., move the bits left or right within the data field. Shifters are useful for many computational applications including arithmetic operations, address generation, variable-length coding, and bit indexing, and have become essential to modern integrated circuit devices such as digital signal processors and general-purpose microprocessors.
- each bit is incrementally shifted one place at a time, so shifting data by n bits would require n clock cycles. This delay is unacceptable for conventional systems having, e.g., 64-bit or 128-bit data values, so an improved design known as a barrel shifter has been devised which can shift long-bit values in a single clock cycle.
- Barrel shifters basically trade chip area of the integrated circuit (number of gates) for speed.
- a barrel shifter relies on a large number of cascaded multiplexers. The minimum number of 2:1 multiplexers required for an n-bit word is n log 2 n; for example, a 64-bit barrel shifter requires 384 multiplexers.
- barrel shifter One common usage of a barrel shifter is in the hardware implementation of floating-point arithmetic. For a floating-point add or subtract operation, the significands of the two numbers must be aligned, which requires shifting the smaller number to the right and increasing its exponent until it matches the exponent of the larger number. This operation is conveniently accomplished by subtracting the exponents, and using the barrel shifter to shift the smaller number to the right by that difference, in one cycle.
- FIG. 1 shows the basic structure of a conventional 64-bit barrel shifter 10 .
- a set of input latches 12 are used to store the operand data and the shift control data for the next shift operation.
- the shift control data includes a one-bit shift direction signal (shift_right) and a six-bit shift amount signal (shift_amount ⁇ 0:5>) representing a shift of 0-63 places.
- the operand data is passed to a formatter and driver 14 which may perform formatting of the data for the multiplexer cascade, and the shift control data is passed to a invert with shift right circuit 16 which generates decoder signals that are later decoded to control the multiplexers.
- the multiplexer cascade has two stages wherein an intermediate or coarsely-shifted value (the operand data shifted by some integer multiple of 8) is selected by a first selector 18 , and then a fully shifted value (further shifted by 0-7) is selected by a second selector 20 using the output of selector 18 .
- Selector 18 selects one group out of eight groups, each having 72 bits, while selector 20 selects one group out of nine groups, each having 64 bits, due to the encoding inherent in the use of invert with shift right circuit 16 .
- the first set of groups have an extra 8 wrap-around bits appended to the operand data to avoid the data transfer delay penalty incurred by long wiring that would otherwise be required.
- Invert with shift right circuit 16 generates a decode signal which may be considered as having two halves, a most-significant bit (MSB) half and a least significant bit (LSB) half.
- the MSB decode bits are used to derive the first selection signal and the LSB decode bits are used to derive the second selection signal, along with the shift_right signal.
- a first decoder 22 computes the selection signal for selector 18
- a second decoder 24 computes the selection signal for selector 20 .
- Decoder 22 is a three-to-eight decoder, i.e., it receives a 3-bit input and computes an 8-bit output
- decoder 24 is a three-to-nine decoder, i.e., it receives a 3-bit input and computes a 9-bit output although it also requires the shift_right input.
- the 64-bit output of selector 20 (raw shifted data) can be further used as an input to a multiplexer and driver 26 to generate final shifted data that is then stored in output latches 28 .
- the present invention is generally directed to a barrel shifter which uses a sign magnitude to 2's complement converter circuit to generate decoder signals for two or more cascaded multiplexer selectors.
- the sign magnitude to 2's complement converter circuit has a sign input which receives an input shift direction and has a magnitude input which receives an input shift amount.
- there are two selectors so the sign magnitude to 2's complement converter circuit generates two decoder signals, and two select signals are derived from the decoder signals using two decoder circuits.
- the sign magnitude to 2's complement converter circuit can compute an output result as a 2's complement of the input shift amount using a shift direction as a sign input, assign a first portion of the output result to the first decoder signal, and assign a second portion of the output result to the second decoder signal.
- the encoding scheme using a sign magnitude to 2's complement converter circuit allows the decoder circuits to be relatively simple, for example, 3-to-8 decoders for an implementation adapted to shift a 64-bit operand value rather than the 4-to-9 decoder required in a conventional barrel shifter.
- the sign magnitude to 2's complement converter circuit can be optimized to generate the decoder signals using no more than three stages of gates.
- the invention also allows for smaller selector logic, e.g., a first selector which selects one group of 71 bits out of eight groups (as opposed to 72 bits in a conventional barrel shifter) and a second selector which selects one group of 64 bits out of eight groups (as opposed to nine groups in a conventional barrel shifter).
- selector logic e.g., a first selector which selects one group of 71 bits out of eight groups (as opposed to 72 bits in a conventional barrel shifter) and a second selector which selects one group of 64 bits out of eight groups (as opposed to nine groups in a conventional barrel shifter).
- FIG. 1 is a block diagram of a conventional 64-bit barrel shifter which uses cascaded multiplexers to provide variable-length shifting in a single execution cycle;
- FIG. 2 is a block diagram of one embodiment of a 64-bit barrel shifter constructed in accordance with the present invention which uses a sign magnitude to 2's complement converter to encode the selector signals;
- FIG. 3 is a block diagram illustrating the inputs and outputs of the sign magnitude to 2's complement converter used in the barrel shifter of FIG. 2 in accordance with one implementation of the present invention
- FIG. 4 is a high-level schematic diagram depicting decoding of the encoded signals from the sign magnitude to 2's complement converter into the selection signals that control the multiplexer cascades in accordance with one implementation of the present invention
- FIG. 5 is a set of converter circuits for the sign magnitude to 2's complement converter in accordance with an exemplary embodiment of the present invention
- FIG. 6 is a chart illustrating the function of the first (higher-order) selector used in the barrel shifter of FIG. 2 in accordance with one implementation of the present invention
- FIG. 7 is a chart illustrating the function of the second (lower-order) selector used in the barrel shifter of FIG. 2 in accordance with one implementation of the present invention
- FIG. 8 is a block diagram showing the front end for an alternative implementation of a barrel shifter in accordance with one implementation of the present invention.
- FIG. 9 is a block diagram of an application for the barrel shifter of FIG. 2 in conjunction with an arithmetic logic unit and other execution units of an integrated circuit microprocessor in accordance with one implementation of the present invention.
- Barrel shifters are extremely valuable for performing variable bit-length shift operations, particularly in execution units of a microprocessor.
- the period of signal transfer along the horizontal bit direction increases, which can make the processing time of this component considerably large.
- cascaded stage selectors can reduce this processing time, they considerably increase the complexity of the decoders and selectors.
- the first selector 18 is relatively simple since it is a straightforward computation to decode a 3-bit signal into an 8-bit signal, but the second selector 20 requires a 9-to-1 selection that leads to circuit inefficiencies.
- the LSB decoder 24 must use both the shift direction signal and the decode signal from invert with shift right circuit 16 in a 4-bit adder circuit, and then use the output of the 4-bit adder circuit as an input to a 4-to-9 decoder circuit. This extra layer of circuitry inserts additional delay into the shifter at a timing critical block (the LSB decoder 24 ).
- the present invention achieves these benefits by replacing the invert with shift right circuit with a sign magnitude to 2's complement converter, which leads to more efficient encoding of the selection signals.
- the design of the sign magnitude to 2's complement converter can further be optimized to avoid any increase of the delay.
- Barrel shifter 30 uses a set of input latches 32 to store the operand data and the shift control data for the next shift operation.
- the shift control data includes the one-bit shift direction signal (shift_right) and the six-bit shift amount signal (shift_amount ⁇ 0:5>) representing a shift of 0-63 places.
- Operand data is passed to a formatter and driver 34 which may perform formatting of the data, and the shift control data is passed to a sign magnitude to 2's complement converter circuit 36 which generates decoder signals that are later decoded to control the cascaded multiplexers.
- Formatter and driver 34 can be used for an operation which is unrelated to the shifting function of the present invention, that is, for concatenating a 32-bit value to yield a 64-bit value according to a particular processor specification.
- sign magnitude to 2's complement converter circuit 36 has a 6-bit magnitude input which receives the shift amount signal, and has a single bit sign input which receives the shift direction signal, i.e., the sign magnitude to 2's complement converter circuit uses the shift_right signal in lieu of the sign bit (the shift_right signal is active, or logical high, to indicate a shift to the right direction, and is inactive, or logical low, to indicate a shift to the left direction).
- the computed result is a 6-bit output (the 2's complement of the shift amount signal); the first three (the most significant) bits of this output are assigned to the decoder signal (shift_msb ⁇ 0:2>) for a first decoder 38 , and the last (the least significant) three bits of this output are assigned to the decoder signal (shift_lsb ⁇ 0:2>) for a second decoder 40 .
- the present invention enables a more efficient encoding scheme which requires the decoders 38 , 40 to be only simple 3-to-8 decoders, i.e., the two decoders 38 , 40 are the same design.
- Decoder 38 derives a first selection signal (s1_sel ⁇ 0:7>) to control a first (higher-order) selector 42
- decoder 40 derives a second selection signal (s2_sel ⁇ 0:7>) to control a second (lower-order) selector 44 .
- s1_sel ⁇ 0:7> to control a first (higher-order) selector 42
- decoder 40 derives a second selection signal (s2_sel ⁇ 0:7>) to control a second (lower-order) selector 44 .
- s1_sel ⁇ 0:7> to control a first (higher-order) selector 42
- decoder 40 derives a second selection signal (s2_s
- selector 42 selects one group out of eight groups, each having 71 bits
- selector 44 selects one group out of eight groups, each having 64 bits, due to the encoding inherent in the use of the sign magnitude to 2's complement converter 36 . Since selector 44 uses only eight input groups (instead of nine), LSB decoder 40 can be greatly simplified over the prior art LSB decoder 24 used in the prior art barrel shifter, leading to significant timing improvement. This approach also reduces the number of transfer gate multiplexers (TG-muxes) for selectors 42 , 44 .
- TG-muxes transfer gate multiplexers
- a transfer gate consists of a pair of coupled transistors, one a p-type metal-oxide semiconducting field effect transistor (pMOSFET) and the other an a n-type metal-oxide semiconducting field effect transistor (nMOSFET).
- the prior art second stage selector is not suited for implementation with a TG-mux because the number of input ports are nine or more. In the TG-mux, only one input port is selected to transfer the data to the output port, so the driver in the selected path has to drive the parasitic capacitance and resistance load for all of the input ports including unselected ports. Driving such a large load with one input driver is especially not suited for advanced technologies which may use a low power supply, e.g., 1.0 volts or less. However, with eight or fewer input ports the use of a TG-mux with such advanced technologies becomes feasible. This scheme additionally increases the noise immunity for the selectors.
- FIG. 5 illustrates a set of converter circuits that can be used to implement the sign magnitude to 2's complement converter.
- the sign bit is zero (in this implementation, the shift_right signal)
- the input value A A 5 , A 4 , . . . A 0
- the sign bit is one (shift_right is high and activated)
- a computed value B B 5 , B 4 , . . . B 0
- Circuit 50 is used to generate the first bit B 0 , and is comprised of two serially connected inverters whose input is A 0 .
- Circuit 52 is used to generate the second bit B 1 , and is comprised of an exclusive OR gate whose inputs are A 0 and A 1 .
- Circuit 54 is a generalized schematic for generating higher bits of the 2's complement.
- An exclusive NOR gate receives the highest place bit as an input along with the output of a NOR gate whose inputs are the remaining lower place bits.
- Circuit 56 is a preferred implementation for generating the sixth bit B 5 .
- the first two bits of the input A 0 and A 1 are inputs to a first NOR gate, and the next three bits A 2 , A 3 , A 4 are inputs to a second NOR gate, and the outputs of these two NOR gates are combined in a NAND gate.
- the output of this NAND gate is combined with the final input bit A 5 in an exclusive OR gate to generate the final 2's complement bit B 5 .
- sign magnitude to 2's complement converter circuit 36 can be implemented with a maximum of three stages of gates, advantageously minimizing the associated delay for generating the decoder signals.
- Table 1 is a truth table showing the decoder outputs according to this scheme.
- both the msb and lsb decoder outputs range from 0 to 7 for both the “shift left” and “shift right” cases which requires only 8-to-1 selection for both selectors 42 , 44 .
- the truth table for the decoders of the prior art barrel shifter seen in FIG. 1 provides 9-to-1 selection for the second selector stage, i.e., output values ranging from 0 to 8.
- FIGS. 6 and 7 depict the function of selectors 42 and 44 , respectively.
- Selector 42 receives the 64-bit pre-shifted operand data from formatter 34 , and as seen in FIG. 6 creates eight different groups (s1g0 through s1g7) each containing 71 bits, including the original 64 bits and the left side 7 bits appended to its right side as wrap-around bits (again provided to avoid the data transfer delay penalty incurred by long wiring that would otherwise be required).
- the multiplexer array within selector 42 provides for intermediate shifting of each group by some integer multiple of a first shift level (higher-order) amount.
- the first shift level unit amount is 8, so the first group s1g0 is unshifted (i.e., shifted by 0 ⁇ 8 bits) beginning with the most significant bit of the input operand data s 1 — 0 and ending with the last of the wrap around bits s 1 — 6 , the second group slgl is shifted by 8 places (1 ⁇ 8) ranging from s 1 — 8 to s 1 — 14 , etc., until the last (eighth) group is shifted by 56 places (7 ⁇ 8) ranging from s 1 — 56 to s 1 — 62 .
- the output group (s1out) selected by selector 42 based on the first select signal then becomes the input group (s2 in) for selector 44 (71 bits, designated s 2 — 0 to s 2 — 70 ).
- Selector 44 receives the 71-bit data from selector 42 , and as seen in FIG. 7 creates eight different groups (s2g0 through s2g7) each containing 64 bits.
- the multiplexer array within selector 44 provides for refined shifting of each group by some integer multiple of a second shift level (lower-order) amount, which is smaller than the first shift level amount.
- the second shift level unit amount is 1, so the first group s2g0 is again unshifted (0 ⁇ 1) ranging from s 2 — 0 to s 2 — 63 , the second group is shifted by 1 place (1 ⁇ 1) ranging from s 2 — 1 to s 2 — 64 , etc., until the last (eighth) group is shifted by 7 places (7 ⁇ 1) ranging from s 2 — 7 to s 2 — 70 .
- the output group (s2out) selected by selector 44 based on the second select signal then becomes the raw shifted data (rsd) for rotator 30 (64 bits, designated rsd 0 to rsd 63 ).
- the raw shifted data from selector 44 may further be provided as an input to multiplexer and driver 46 to generate final shifted data that can be stored in output latches 48 .
- Multiplexer and driver 46 can be used for another operation which is not directly related to the shifting function of the present invention, that is, for shifting a different operand again according to a particular processor specification; this optional function does not affect the timing of the barrel shifter operation.
- the two selectors 42 , 44 with their corresponding shift level unit amounts (8 and 1) represent different levels of a hierarchical design for implementing a cascaded barrel shifter. Accordingly, other implementations of a barrel shifter based on the present invention may have a different hierarchy in which case the numbers of bits in the decoder signals, the number of selector stages, the number of groups created by the selectors, or the shift level unit amounts may vary from the disclosed embodiments. For example, a 64-bit shifter could alternatively be implemented in three stages each using a two-bit decoder signal, with shift level unit amounts of 16, 4 and 1 (four groups to select from at each selector stage).
- a 512-bit rotator could also be implemented in three stages with each stage using a 3-bit decoder signal and shift level unit amounts of 64, 8 and 1 (eight groups to select from at each selector stage).
- the decoder signals are still generated by a sign magnitude to 2's complement converter.
- FIG. 8 illustrates a further alternative embodiment 30 ′ for a 64-bit barrel shifter which differs from barrel shifter 30 only in the relative placement of the sign magnitude to 2's complement converter circuit 36 and input latches 32 .
- the 2's complement conversion occurs in the previous cycle and the decoder signals shift_msb ⁇ 0:2> and shift_lsb ⁇ 0:2> are placed in the input latches 32 along with the operand data. In this manner pipelined operations within shifter 30 ′ can execute even faster since the encoding stage is placed outside of the execution cycle.
- the barrel shifter of the present invention may be used in a wide variety of applications which require shifting or rotation, and is particularly useful within execution units of a microprocessor.
- One particular application shown in FIG. 9 is an execution unit 60 which includes the barrel shifter 30 , an arithmetic logic unit (ALU) 62 , and a multi-cycle operation block 64 .
- Barrel shifter 30 receives the operand data and the control data from two operand busses and control signal lines, which are also connected to ALU 62 and multi-cycle operation block 64 .
- ALU 62 additionally receives a carry signal as is known in the art.
- ALU 62 is used for logical, arithmetic and compare operations.
- Shifter 30 and ALU 62 are implemented as single-cycle operation blocks because they are often called most frequently from program code. Other operations can be carried out by multi-cycle operation block 64 , such as counting the number of zero or nonzero bits in the operand or a miscellaneous operation defined by a particular instruction set.
- the outputs of shifter 30 , ALU 62 and multi-cycle operation block 64 are connected to the input ports of a 3-way NOR gate 66 , and the outputs are tied to “0” by the control signals when the given component is not selected for operation, so the output of NOR gate 66 is the output of the chosen component 30 , 62 or 64 according to the particular operation being executed.
- Execution unit 60 is only one example for utilizing the barrel shifter of the present invention, and those skilled in the art will appreciate that it may be utilized in many other applications including for example within an ALU instead of outside the ALU as seen in FIG. 9 .
- the present invention accordingly imparts several advantages over prior art barrel shifters.
- the use of the sign magnitude to 2's complement encoding allows not only for faster operation, but also requires less area of the integrated circuit (fewer gates), and thus leads to reduced power consumption. High speed operation can be achieved even for timing critical cases such as used in a one-cycle operation path.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Executing Machine-Instructions (AREA)
- Dc Digital Transmission (AREA)
Abstract
Description
- 1. Field of the Invention
- The present invention generally relates to logic circuits for computational devices, and more particularly to a barrel shifter or rotator.
- 2. Description of the Related Art
- A shifter, or rotator, is a combinational logic circuit that can shift an operand data value by a specified number of places (bits), i.e., move the bits left or right within the data field. Shifters are useful for many computational applications including arithmetic operations, address generation, variable-length coding, and bit indexing, and have become essential to modern integrated circuit devices such as digital signal processors and general-purpose microprocessors.
- In the simplest shifter circuit, each bit is incrementally shifted one place at a time, so shifting data by n bits would require n clock cycles. This delay is unacceptable for conventional systems having, e.g., 64-bit or 128-bit data values, so an improved design known as a barrel shifter has been devised which can shift long-bit values in a single clock cycle. Barrel shifters basically trade chip area of the integrated circuit (number of gates) for speed. A barrel shifter relies on a large number of cascaded multiplexers. The minimum number of 2:1 multiplexers required for an n-bit word is n log2 n; for example, a 64-bit barrel shifter requires 384 multiplexers. One common usage of a barrel shifter is in the hardware implementation of floating-point arithmetic. For a floating-point add or subtract operation, the significands of the two numbers must be aligned, which requires shifting the smaller number to the right and increasing its exponent until it matches the exponent of the larger number. This operation is conveniently accomplished by subtracting the exponents, and using the barrel shifter to shift the smaller number to the right by that difference, in one cycle.
-
FIG. 1 shows the basic structure of a conventional 64-bit barrel shifter 10. A set ofinput latches 12 are used to store the operand data and the shift control data for the next shift operation. The shift control data includes a one-bit shift direction signal (shift_right) and a six-bit shift amount signal (shift_amount<0:5>) representing a shift of 0-63 places. The operand data is passed to a formatter anddriver 14 which may perform formatting of the data for the multiplexer cascade, and the shift control data is passed to a invert with shiftright circuit 16 which generates decoder signals that are later decoded to control the multiplexers. In this design, the multiplexer cascade has two stages wherein an intermediate or coarsely-shifted value (the operand data shifted by some integer multiple of 8) is selected by afirst selector 18, and then a fully shifted value (further shifted by 0-7) is selected by asecond selector 20 using the output ofselector 18.Selector 18 selects one group out of eight groups, each having 72 bits, whileselector 20 selects one group out of nine groups, each having 64 bits, due to the encoding inherent in the use of invert with shiftright circuit 16. The first set of groups have an extra 8 wrap-around bits appended to the operand data to avoid the data transfer delay penalty incurred by long wiring that would otherwise be required. Invert with shiftright circuit 16 generates a decode signal which may be considered as having two halves, a most-significant bit (MSB) half and a least significant bit (LSB) half. The MSB decode bits are used to derive the first selection signal and the LSB decode bits are used to derive the second selection signal, along with the shift_right signal. Afirst decoder 22 computes the selection signal forselector 18, and asecond decoder 24 computes the selection signal forselector 20.Decoder 22 is a three-to-eight decoder, i.e., it receives a 3-bit input and computes an 8-bit output, whiledecoder 24 is a three-to-nine decoder, i.e., it receives a 3-bit input and computes a 9-bit output although it also requires the shift_right input. The 64-bit output of selector 20 (raw shifted data) can be further used as an input to a multiplexer anddriver 26 to generate final shifted data that is then stored inoutput latches 28. - The present invention is generally directed to a barrel shifter which uses a sign magnitude to 2's complement converter circuit to generate decoder signals for two or more cascaded multiplexer selectors. The sign magnitude to 2's complement converter circuit has a sign input which receives an input shift direction and has a magnitude input which receives an input shift amount. In the exemplary embodiment there are two selectors, so the sign magnitude to 2's complement converter circuit generates two decoder signals, and two select signals are derived from the decoder signals using two decoder circuits. The sign magnitude to 2's complement converter circuit can compute an output result as a 2's complement of the input shift amount using a shift direction as a sign input, assign a first portion of the output result to the first decoder signal, and assign a second portion of the output result to the second decoder signal. The encoding scheme using a sign magnitude to 2's complement converter circuit allows the decoder circuits to be relatively simple, for example, 3-to-8 decoders for an implementation adapted to shift a 64-bit operand value rather than the 4-to-9 decoder required in a conventional barrel shifter. For that implementation, the sign magnitude to 2's complement converter circuit can be optimized to generate the decoder signals using no more than three stages of gates. The invention also allows for smaller selector logic, e.g., a first selector which selects one group of 71 bits out of eight groups (as opposed to 72 bits in a conventional barrel shifter) and a second selector which selects one group of 64 bits out of eight groups (as opposed to nine groups in a conventional barrel shifter).
- The above as well as additional objectives, features, and advantages of the present invention will become apparent in the following detailed written description.
- The present invention may be better understood, and its numerous objects, features, and advantages made apparent to those skilled in the art by referencing the accompanying drawings.
-
FIG. 1 is a block diagram of a conventional 64-bit barrel shifter which uses cascaded multiplexers to provide variable-length shifting in a single execution cycle; -
FIG. 2 is a block diagram of one embodiment of a 64-bit barrel shifter constructed in accordance with the present invention which uses a sign magnitude to 2's complement converter to encode the selector signals; -
FIG. 3 is a block diagram illustrating the inputs and outputs of the sign magnitude to 2's complement converter used in the barrel shifter ofFIG. 2 in accordance with one implementation of the present invention; -
FIG. 4 is a high-level schematic diagram depicting decoding of the encoded signals from the sign magnitude to 2's complement converter into the selection signals that control the multiplexer cascades in accordance with one implementation of the present invention; -
FIG. 5 is a set of converter circuits for the sign magnitude to 2's complement converter in accordance with an exemplary embodiment of the present invention; -
FIG. 6 is a chart illustrating the function of the first (higher-order) selector used in the barrel shifter ofFIG. 2 in accordance with one implementation of the present invention; -
FIG. 7 is a chart illustrating the function of the second (lower-order) selector used in the barrel shifter ofFIG. 2 in accordance with one implementation of the present invention; -
FIG. 8 is a block diagram showing the front end for an alternative implementation of a barrel shifter in accordance with one implementation of the present invention; and -
FIG. 9 is a block diagram of an application for the barrel shifter ofFIG. 2 in conjunction with an arithmetic logic unit and other execution units of an integrated circuit microprocessor in accordance with one implementation of the present invention. - The use of the same reference symbols in different drawings indicates similar or identical items.
- Barrel shifters (rotators) are extremely valuable for performing variable bit-length shift operations, particularly in execution units of a microprocessor. However, as the number of bits being shifted increases, the period of signal transfer along the horizontal bit direction increases, which can make the processing time of this component considerably large. While cascaded stage selectors can reduce this processing time, they considerably increase the complexity of the decoders and selectors. For example, in the prior
art barrel shifter 10 seen inFIG. 1 , thefirst selector 18 is relatively simple since it is a straightforward computation to decode a 3-bit signal into an 8-bit signal, but thesecond selector 20 requires a 9-to-1 selection that leads to circuit inefficiencies. TheLSB decoder 24 must use both the shift direction signal and the decode signal from invert with shiftright circuit 16 in a 4-bit adder circuit, and then use the output of the 4-bit adder circuit as an input to a 4-to-9 decoder circuit. This extra layer of circuitry inserts additional delay into the shifter at a timing critical block (the LSB decoder 24). - It would, therefore, be desirable to devise an improved barrel shifter which could minimize or reduce the delay in the timing-critical decoder circuits. It would be further advantageous if the method could allow for a reduction in the number of selector inputs. The present invention achieves these benefits by replacing the invert with shift right circuit with a sign magnitude to 2's complement converter, which leads to more efficient encoding of the selection signals. The design of the sign magnitude to 2's complement converter can further be optimized to avoid any increase of the delay. These improvements contribute to faster operation and reduced area and power, as well as noise immunity improvement. As the number of bits to be shifted increases, the effectiveness of this scheme likewise increases.
- With reference now to the figures, and in particular with reference to
FIG. 2 , there is depicted oneembodiment 30 of an improved barrel shifter constructed in accordance with the present invention.Barrel shifter 30 uses a set ofinput latches 32 to store the operand data and the shift control data for the next shift operation. The shift control data includes the one-bit shift direction signal (shift_right) and the six-bit shift amount signal (shift_amount<0:5>) representing a shift of 0-63 places. Operand data is passed to a formatter anddriver 34 which may perform formatting of the data, and the shift control data is passed to a sign magnitude to 2'scomplement converter circuit 36 which generates decoder signals that are later decoded to control the cascaded multiplexers. Formatter anddriver 34 can be used for an operation which is unrelated to the shifting function of the present invention, that is, for concatenating a 32-bit value to yield a 64-bit value according to a particular processor specification. - As seen in
FIG. 3 , sign magnitude to 2'scomplement converter circuit 36 has a 6-bit magnitude input which receives the shift amount signal, and has a single bit sign input which receives the shift direction signal, i.e., the sign magnitude to 2's complement converter circuit uses the shift_right signal in lieu of the sign bit (the shift_right signal is active, or logical high, to indicate a shift to the right direction, and is inactive, or logical low, to indicate a shift to the left direction). The computed result is a 6-bit output (the 2's complement of the shift amount signal); the first three (the most significant) bits of this output are assigned to the decoder signal (shift_msb<0:2>) for afirst decoder 38, and the last (the least significant) three bits of this output are assigned to the decoder signal (shift_lsb<0:2>) for asecond decoder 40. - As seen further in
FIG. 4 , the present invention enables a more efficient encoding scheme which requires thedecoders decoders Decoder 38 derives a first selection signal (s1_sel<0:7>) to control a first (higher-order)selector 42, anddecoder 40 derives a second selection signal (s2_sel<0:7>) to control a second (lower-order)selector 44. In further contrast to the conventional barrel shifter illustrated inFIG. 1 ,selector 42 selects one group out of eight groups, each having 71 bits, andselector 44 selects one group out of eight groups, each having 64 bits, due to the encoding inherent in the use of the sign magnitude to 2'scomplement converter 36. Sinceselector 44 uses only eight input groups (instead of nine),LSB decoder 40 can be greatly simplified over the priorart LSB decoder 24 used in the prior art barrel shifter, leading to significant timing improvement. This approach also reduces the number of transfer gate multiplexers (TG-muxes) forselectors -
FIG. 5 illustrates a set of converter circuits that can be used to implement the sign magnitude to 2's complement converter. For general sign magnitude to 2's complement conversion, if the sign bit is zero (in this implementation, the shift_right signal), then the input value A (A5, A4, . . . A0) is used as the output of the sign magnitude to 2's complement converter. But if the sign bit is one (shift_right is high and activated), a computed value B (B5, B4, . . . B0) is used as the output of the sign magnitude to 2's complement converter. Accordingly, the converter circuits shown inFIG. 5 are only necessary for this latter circumstance, i.e., shift_right=1, and if shift_right=0 then sign magnitude to 2'scomplement converter circuit 36 simply passes the shift amount input as the 2's complement output.Circuit 50 is used to generate the first bit B0, and is comprised of two serially connected inverters whose input is A0. Circuit 52 is used to generate the second bit B1, and is comprised of an exclusive OR gate whose inputs are A0 and A1. Circuit 54 is a generalized schematic for generating higher bits of the 2's complement. An exclusive NOR gate receives the highest place bit as an input along with the output of a NOR gate whose inputs are the remaining lower place bits.Circuit 56 is a preferred implementation for generating the sixth bit B5. In this construction, the first two bits of the input A0 and A1 are inputs to a first NOR gate, and the next three bits A2, A3, A4 are inputs to a second NOR gate, and the outputs of these two NOR gates are combined in a NAND gate. The output of this NAND gate is combined with the final input bit A5 in an exclusive OR gate to generate the final 2's complement bit B5. Thus, sign magnitude to 2'scomplement converter circuit 36 can be implemented with a maximum of three stages of gates, advantageously minimizing the associated delay for generating the decoder signals. Table 1 is a truth table showing the decoder outputs according to this scheme. -
TABLE 1 shift left (shift_right = 0) shift right (shift_right = 1) msb decoder lsb decoder msb decoder lsb decoder shift amount in out in out in out in out 000000 000 0 000 0 000 0 000 0 000001 000 0 001 1 111 7 111 7 000010 000 0 010 2 111 7 110 6 . . . . . . . . . . . . . . . 000101 000 0 101 5 111 7 011 3 000110 000 0 110 6 111 7 010 2 000111 000 0 111 7 111 7 001 1 001000 001 1 000 0 111 7 000 0 001001 001 1 001 1 110 6 111 7 001010 001 1 010 2 110 6 110 6 . . . . . . . . . . . . . . . 111101 111 7 101 5 000 0 011 3 111110 111 7 110 6 000 0 010 2 111111 111 7 111 7 000 0 001 1
According to the above table, both the msb and lsb decoder outputs range from 0 to 7 for both the “shift left” and “shift right” cases which requires only 8-to-1 selection for bothselectors FIG. 1 provides 9-to-1 selection for the second selector stage, i.e., output values ranging from 0 to 8. -
FIGS. 6 and 7 depict the function ofselectors Selector 42 receives the 64-bit pre-shifted operand data fromformatter 34, and as seen inFIG. 6 creates eight different groups (s1g0 through s1g7) each containing 71 bits, including the original 64 bits and theleft side 7 bits appended to its right side as wrap-around bits (again provided to avoid the data transfer delay penalty incurred by long wiring that would otherwise be required). The multiplexer array withinselector 42 provides for intermediate shifting of each group by some integer multiple of a first shift level (higher-order) amount. In this implementation, the first shift level unit amount is 8, so the first group s1g0 is unshifted (i.e., shifted by 0×8 bits) beginning with the most significant bit of the input operand data s1— 0 and ending with the last of the wrap around bits s1— 6, the second group slgl is shifted by 8 places (1×8) ranging from s1— 8 to s1— 14, etc., until the last (eighth) group is shifted by 56 places (7×8) ranging from s1— 56 to s1— 62. The output group (s1out) selected byselector 42 based on the first select signal then becomes the input group (s2 in) for selector 44 (71 bits, designated s2— 0 to s2— 70). -
Selector 44 receives the 71-bit data fromselector 42, and as seen inFIG. 7 creates eight different groups (s2g0 through s2g7) each containing 64 bits. The multiplexer array withinselector 44 provides for refined shifting of each group by some integer multiple of a second shift level (lower-order) amount, which is smaller than the first shift level amount. In this implementation, the second shift level unit amount is 1, so the first group s2g0 is again unshifted (0×1) ranging from s2— 0 to s2— 63, the second group is shifted by 1 place (1×1) ranging from s2— 1 to s2— 64, etc., until the last (eighth) group is shifted by 7 places (7×1) ranging from s2— 7 to s2— 70. The output group (s2out) selected byselector 44 based on the second select signal then becomes the raw shifted data (rsd) for rotator 30 (64 bits, designated rsd0 to rsd63). The raw shifted data fromselector 44 may further be provided as an input to multiplexer anddriver 46 to generate final shifted data that can be stored in output latches 48. Multiplexer anddriver 46 can be used for another operation which is not directly related to the shifting function of the present invention, that is, for shifting a different operand again according to a particular processor specification; this optional function does not affect the timing of the barrel shifter operation. - Those skilled in the art will recognize that the two
selectors implementation 30 of a barrel shift are exemplary only, and should not be construed in a limiting sense. -
FIG. 8 illustrates a furtheralternative embodiment 30′ for a 64-bit barrel shifter which differs frombarrel shifter 30 only in the relative placement of the sign magnitude to 2'scomplement converter circuit 36 and input latches 32. In this alternative embodiment the 2's complement conversion occurs in the previous cycle and the decoder signals shift_msb<0:2> and shift_lsb<0:2> are placed in the input latches 32 along with the operand data. In this manner pipelined operations withinshifter 30′ can execute even faster since the encoding stage is placed outside of the execution cycle. - The barrel shifter of the present invention may be used in a wide variety of applications which require shifting or rotation, and is particularly useful within execution units of a microprocessor. One particular application shown in
FIG. 9 is anexecution unit 60 which includes thebarrel shifter 30, an arithmetic logic unit (ALU) 62, and amulti-cycle operation block 64.Barrel shifter 30 receives the operand data and the control data from two operand busses and control signal lines, which are also connected to ALU 62 andmulti-cycle operation block 64.ALU 62 additionally receives a carry signal as is known in the art.ALU 62 is used for logical, arithmetic and compare operations.Shifter 30 andALU 62 are implemented as single-cycle operation blocks because they are often called most frequently from program code. Other operations can be carried out bymulti-cycle operation block 64, such as counting the number of zero or nonzero bits in the operand or a miscellaneous operation defined by a particular instruction set. The outputs ofshifter 30,ALU 62 andmulti-cycle operation block 64 are connected to the input ports of a 3-way NORgate 66, and the outputs are tied to “0” by the control signals when the given component is not selected for operation, so the output of NORgate 66 is the output of the chosencomponent Execution unit 60 is only one example for utilizing the barrel shifter of the present invention, and those skilled in the art will appreciate that it may be utilized in many other applications including for example within an ALU instead of outside the ALU as seen inFIG. 9 . - The present invention accordingly imparts several advantages over prior art barrel shifters. The use of the sign magnitude to 2's complement encoding allows not only for faster operation, but also requires less area of the integrated circuit (fewer gates), and thus leads to reduced power consumption. High speed operation can be achieved even for timing critical cases such as used in a one-cycle operation path. These benefits of the present invention are further amplified as the number of bits increases.
- Although the invention has been described with reference to specific embodiments, this description is not meant to be construed in a limiting sense. Various modifications of the disclosed embodiments, as well as alternative embodiments of the invention, will become apparent to persons skilled in the art upon reference to the description of the invention. It is therefore contemplated that such modifications can be made without departing from the spirit or scope of the present invention as defined in the appended claims.
Claims (25)
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/538,001 US9021000B2 (en) | 2012-06-29 | 2012-06-29 | High speed and low power circuit structure for barrel shifter |
GB1500098.7A GB2518317B (en) | 2012-06-29 | 2013-04-08 | High speed and low power circuit structure for barrel shifter |
PCT/JP2013/002394 WO2014002342A1 (en) | 2012-06-29 | 2013-04-08 | High speed and low power circuit structure for barrel shifter |
US14/094,794 US9501260B2 (en) | 2012-06-29 | 2013-12-03 | High speed and low power circuit structure for barrel shifter |
US14/611,223 US9996317B2 (en) | 2012-06-29 | 2015-01-31 | High speed and low power circuit structure for barrel shifter |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/538,001 US9021000B2 (en) | 2012-06-29 | 2012-06-29 | High speed and low power circuit structure for barrel shifter |
Related Child Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/094,794 Continuation US9501260B2 (en) | 2012-06-29 | 2013-12-03 | High speed and low power circuit structure for barrel shifter |
US14/611,223 Continuation US9996317B2 (en) | 2012-06-29 | 2015-01-31 | High speed and low power circuit structure for barrel shifter |
Publications (2)
Publication Number | Publication Date |
---|---|
US20140006466A1 true US20140006466A1 (en) | 2014-01-02 |
US9021000B2 US9021000B2 (en) | 2015-04-28 |
Family
ID=49779296
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/538,001 Expired - Fee Related US9021000B2 (en) | 2012-06-29 | 2012-06-29 | High speed and low power circuit structure for barrel shifter |
US14/094,794 Active 2033-08-09 US9501260B2 (en) | 2012-06-29 | 2013-12-03 | High speed and low power circuit structure for barrel shifter |
US14/611,223 Active 2033-03-27 US9996317B2 (en) | 2012-06-29 | 2015-01-31 | High speed and low power circuit structure for barrel shifter |
Family Applications After (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/094,794 Active 2033-08-09 US9501260B2 (en) | 2012-06-29 | 2013-12-03 | High speed and low power circuit structure for barrel shifter |
US14/611,223 Active 2033-03-27 US9996317B2 (en) | 2012-06-29 | 2015-01-31 | High speed and low power circuit structure for barrel shifter |
Country Status (3)
Country | Link |
---|---|
US (3) | US9021000B2 (en) |
GB (1) | GB2518317B (en) |
WO (1) | WO2014002342A1 (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109756231A (en) * | 2018-12-27 | 2019-05-14 | 北京思朗科技有限责任公司 | Cyclic shift processing unit and method |
CN110837510A (en) * | 2019-11-13 | 2020-02-25 | 北京深演智能科技股份有限公司 | Data processing method, device, storage medium and processor |
CN117573065A (en) * | 2024-01-17 | 2024-02-20 | 青岛本原微电子有限公司 | Multifunctional shift operation device and operation method for processor |
WO2024087726A1 (en) * | 2022-10-26 | 2024-05-02 | Huawei Technologies Co., Ltd. | Energetically efficient representation of signed integer non uniform weights |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9607682B1 (en) * | 2016-03-28 | 2017-03-28 | Amazon Technologies, Inc. | Address decoding circuit |
WO2022235517A2 (en) * | 2021-05-05 | 2022-11-10 | Uniquify, Inc. | Implementations and methods for processing neural network in semiconductor hardware |
JP7506276B2 (en) * | 2021-05-05 | 2024-06-25 | ユニークファイ, インク | Implementations and methods for processing neural networks in semiconductor hardware - Patents.com |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4785393A (en) * | 1984-07-09 | 1988-11-15 | Advanced Micro Devices, Inc. | 32-Bit extended function arithmetic-logic unit on a single chip |
US4825400A (en) * | 1986-01-13 | 1989-04-25 | General Electric Company | Floating point accumulator circuit |
US5124583A (en) * | 1990-02-09 | 1992-06-23 | Hitachi, Ltd. | Dc-powered josephson integrated circuit |
US5272654A (en) * | 1991-09-26 | 1993-12-21 | Advanced Micro Devices, Inc. | System for converting a floating point signed magnitude binary number to a two's complement binary number |
US7395294B1 (en) * | 2003-01-10 | 2008-07-01 | Altera Corporation | Arithmetic logic unit |
US7415542B2 (en) * | 2004-06-18 | 2008-08-19 | Analog Devices, Inc. | Micro-programmable filter engine having plurality of filter elements interconnected in chain configuration wherein engine supports multiple filters from filter elements |
Family Cites Families (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS6132139A (en) | 1984-07-24 | 1986-02-14 | Nec Corp | Two-way barrel shift circuit |
JPH01161433A (en) | 1987-12-17 | 1989-06-26 | Nec Corp | Rotating circuit |
JPH0277933A (en) | 1988-09-14 | 1990-03-19 | Matsushita Electric Ind Co Ltd | Data transfer device |
JPH02178831A (en) | 1988-12-29 | 1990-07-11 | Nec Home Electron Ltd | Circuit and method for multiplying data by-1 |
JPH0421025A (en) | 1990-05-15 | 1992-01-24 | Toshiba Corp | Left and right shifter |
JPH04245323A (en) | 1991-01-30 | 1992-09-01 | Fujitsu Ltd | Barrel shifter circuit |
US5555202A (en) | 1994-12-05 | 1996-09-10 | Cirrus Logic, Inc. | Low-power, high-performance barrel shifter |
JP3436994B2 (en) | 1994-12-05 | 2003-08-18 | 三菱電機株式会社 | Shift device |
JP3311624B2 (en) | 1995-12-29 | 2002-08-05 | ルーセント テクノロジーズ インコーポレイテッド | Barrel shifter control method and integrated circuit |
JP5182053B2 (en) | 2008-12-10 | 2013-04-10 | 日本電気株式会社 | Barrel shifter device and barrel shift method |
-
2012
- 2012-06-29 US US13/538,001 patent/US9021000B2/en not_active Expired - Fee Related
-
2013
- 2013-04-08 WO PCT/JP2013/002394 patent/WO2014002342A1/en active Application Filing
- 2013-04-08 GB GB1500098.7A patent/GB2518317B/en active Active
- 2013-12-03 US US14/094,794 patent/US9501260B2/en active Active
-
2015
- 2015-01-31 US US14/611,223 patent/US9996317B2/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4785393A (en) * | 1984-07-09 | 1988-11-15 | Advanced Micro Devices, Inc. | 32-Bit extended function arithmetic-logic unit on a single chip |
US4825400A (en) * | 1986-01-13 | 1989-04-25 | General Electric Company | Floating point accumulator circuit |
US5124583A (en) * | 1990-02-09 | 1992-06-23 | Hitachi, Ltd. | Dc-powered josephson integrated circuit |
US5272654A (en) * | 1991-09-26 | 1993-12-21 | Advanced Micro Devices, Inc. | System for converting a floating point signed magnitude binary number to a two's complement binary number |
US7395294B1 (en) * | 2003-01-10 | 2008-07-01 | Altera Corporation | Arithmetic logic unit |
US7415542B2 (en) * | 2004-06-18 | 2008-08-19 | Analog Devices, Inc. | Micro-programmable filter engine having plurality of filter elements interconnected in chain configuration wherein engine supports multiple filters from filter elements |
Non-Patent Citations (1)
Title |
---|
Fairchild Semiconductor Corporation, NCSZ27 TinyLogic (R) UHS 3-Input NOR Gate, September 2004 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109756231A (en) * | 2018-12-27 | 2019-05-14 | 北京思朗科技有限责任公司 | Cyclic shift processing unit and method |
CN110837510A (en) * | 2019-11-13 | 2020-02-25 | 北京深演智能科技股份有限公司 | Data processing method, device, storage medium and processor |
WO2024087726A1 (en) * | 2022-10-26 | 2024-05-02 | Huawei Technologies Co., Ltd. | Energetically efficient representation of signed integer non uniform weights |
CN117573065A (en) * | 2024-01-17 | 2024-02-20 | 青岛本原微电子有限公司 | Multifunctional shift operation device and operation method for processor |
Also Published As
Publication number | Publication date |
---|---|
US9021000B2 (en) | 2015-04-28 |
US9996317B2 (en) | 2018-06-12 |
WO2014002342A1 (en) | 2014-01-03 |
US20150149518A1 (en) | 2015-05-28 |
GB2518317A (en) | 2015-03-18 |
US9501260B2 (en) | 2016-11-22 |
GB2518317B (en) | 2020-10-28 |
US20140089363A1 (en) | 2014-03-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9996317B2 (en) | High speed and low power circuit structure for barrel shifter | |
EP0827069B1 (en) | Arithmetic circuit and method | |
US7308470B2 (en) | Smaller and lower power static mux circuitry in generating multiplier partial product signals | |
US6970897B2 (en) | Self-timed transmission system and method for processing multiple data sets | |
US5636157A (en) | Modular 64-bit integer adder | |
US7024445B2 (en) | Method and apparatus for use in booth-encoded multiplication | |
US6183122B1 (en) | Multiplier sign extension | |
Ykuntam et al. | Design of 32-bit carry select adder with reduced area | |
Vun et al. | Thermometer code based modular arithmetic | |
US5777918A (en) | Fast multiple operands adder/subtracter based on shifting | |
GB2127187A (en) | Circuits for operating on N- digit operands | |
US5103420A (en) | Method and apparatus for srt division using gray coded quotient bit signals | |
Singh et al. | Modified booth multiplier with carry select adder using 3-stage pipelining technique | |
US7159004B2 (en) | Adder, multiplier and integrated circuit | |
JPH09185493A (en) | Integrated circuit for adder | |
JP2004265204A (en) | Carry look-ahead circuit and adder circuit using it | |
ANUSHA et al. | A Modified Partial Product Generator for Redundant Binary Multipliers | |
MOUNIKA et al. | Scalable Digital CMOS Comparator using a Parallel Prefix Tree | |
FAJOOLUNISSA et al. | Design and Implementation of High Speed Carry Skip Adder | |
Jui et al. | Efficient algorithm and hardware implementation of 3N for arithmetic and for radix-8 encodings | |
US9904511B2 (en) | High performance shifter circuit | |
JP3790534B2 (en) | Logic circuit | |
Archanadevi et al. | Design of a dynamic CMOS incrementer/decrementer and a parallel cascading architecture | |
JPH04227336A (en) | Decoder | |
JP2000269805A (en) | Logic circuit |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:YASUDA, TAKEO;REEL/FRAME:028470/0953 Effective date: 20120629 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FEPP | Fee payment procedure |
Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
LAPS | Lapse for failure to pay maintenance fees |
Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |
|
FP | Expired due to failure to pay maintenance fee |
Effective date: 20190428 |